huggingface/text-generation-inference
5
0
Fork 0
mirror of https://github.com/huggingface/text-generation-inference.git synced 2025-09-11 20:34:54 +00:00
Code Issues Packages Projects Releases Wiki Activity

fix: setting the rotary base from the config for the grouped query models.

Browse Source
This commit is contained in:
Nilabhra 2024-05-14 10:14:18 +04:00
parent 5e8db7c14f
commit 9e4b25c66b
1 changed files with 6 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
server/text_generation_server/models/custom_modeling/flash_rw_modeling.py
View File

@ -6,12 +6,14 @@ from torch import nn
from transformers.configuration_utils import PretrainedConfig from transformers.configuration_utils import PretrainedConfig
from transformers.modeling_utils import PreTrainedModel from transformers.modeling_utils import PreTrainedModel
from text_generation_server.utils import paged_attention, flash_attn from text_generation_server.utils import flash_attn, paged_attention
from text_generation_server.utils.layers import ( from text_generation_server.utils.layers import (
TensorParallelRowLinear, FastLayerNorm,
PositionRotaryEmbedding,
SpeculativeHead,
TensorParallelColumnLinear, TensorParallelColumnLinear,
TensorParallelEmbedding, TensorParallelEmbedding,
SpeculativeHead, TensorParallelRowLinear,
get_linear, get_linear,
) )
from text_generation_server.layers.layernorm import ( from text_generation_server.layers.layernorm import (
@ -479,6 +481,7 @@ class FlashRWLayer(nn.Module):
return mlp_output, residual return mlp_output, residual
class FlashRWLayerNorm(nn.Module): class FlashRWLayerNorm(nn.Module):
def __init__(self, config, prefix, weights): def __init__(self, config, prefix, weights):
super().__init__() super().__init__()